Device selection
Confirm that the browser is listening to the headset, USB microphone or built-in input you actually intend to use.
Choose an input device, speak normally and check what the browser receives. Use delayed replay, waveform, spectrum and level readings to spot silence, background noise or clipping.
Microphone access is requested only after you start the test. Render any local-only or no-recording claim only after confirming the implementation.
State exactly what the source code does. If audio frames remain in memory and are neither recorded nor sent to a server, say so. If analytics, diagnostics, recording or remote processing touches microphone data, disclose that instead. Stopping the test should release the media tracks.
Input Level
-∞dBFS
Peak
-∞dBFS
Noise floor
-100dBFS
Replay delay
1.0s
Microphone Tester opens a live audio loop from your input device and replays it back through the speakers after a delay you control. The page shows the waveform, frequency spectrum, and peak meter alongside running statistics for input level, peak, noise floor, detected speech, and clipping. Everything runs locally — the audio buffer never leaves your browser.
Use it before joining a meeting, recording a podcast, or troubleshooting a quiet headset. The replay confirms the right device is selected and that the signal is reaching the operating system.
Speech usually peaks around -12 dBFS to -6 dBFS. A noise floor below -50 dBFS is healthy for most rooms. Clipping above 0% means the input is too hot — back off the gain or move farther from the mic.
The replay path is loud enough to feed back into the microphone. Wear headphones, lower the volume, or disable Monitor before raising speakers.
Audio frames are read by the browser and processed in this tab. Nothing is recorded to disk and nothing is uploaded. Closing the tab releases the device immediately.
Select the intended input, allow access and speak at your normal distance. A moving level meter confirms that a signal is arriving; delayed replay lets you hear noise, distortion, dropouts and the wrong device before using another app.
Confirm that the browser is listening to the headset, USB microphone or built-in input you actually intend to use.
Waveform, spectrum and meter views show different aspects of level and frequency content without replacing a proper calibration tool.
Hear the captured signal after a short delay so the microphone is less likely to feed directly back into the speakers.
Average level, peak, background noise, detected speech and clipping help identify obvious setup problems.
Select the exact microphone you plan to use. Device names may appear only after permission is granted.
Allow microphone access for this site. The test cannot start while permission is blocked.
Use your normal speaking distance and volume rather than shouting into the capsule.
Use headphones for replay, then check noise, peak level, clipping and whether speech is detected consistently.
Adjust input gain, placement, cable, mute state or room noise and repeat the same test conditions.
dBFS is measured relative to the digital maximum: 0 dBFS is the ceiling, while normal signals are negative. Do not publish universal “ideal” numbers without a documented measurement method.
Peak shows the loudest recent sample or frame. Repeated clipping means the input is reaching or exceeding the digital ceiling and should be reduced at the source.
Noise floor estimates the level present when you are not speaking. Fans, air conditioning, electrical hum and high microphone gain can raise it.
Delay controls how long the browser waits before playing the input back. Monitoring can create feedback, so headphones are the safest default.
Waveform shows level over time, spectrum shows frequency energy, and the meter is the quickest way to check whether the signal is too low or too hot.
State exactly what the source code does. If audio frames remain in memory and are neither recorded nor sent to a server, say so. If analytics, diagnostics, recording or remote processing touches microphone data, disclose that instead. Stopping the test should release the media tracks.
Check the selected device, hardware mute, operating-system input level and browser permission. Another app may also have exclusive access.
Some delay is intentional. Heavy tabs, Bluetooth devices and high system load can add latency or dropouts.
Reduce microphone gain, move farther away and verify that clipping falls to zero. Digital clipping cannot be repaired by turning playback volume down.
That is acoustic feedback. Stop monitoring, use headphones or reduce speaker volume immediately.
Only claim “no recording” if the implementation never writes or persists microphone audio. A live test can still read audio frames in memory to display levels and replay the signal.
Browsers often hide device labels until permission is granted. Allow access, then reopen the device list.
Check Monitor, replay volume, mute state and the output device. Use headphones and verify that the page is not blocked from autoplay.
Clipping means the input reached the digital ceiling and the waveform was cut off. Lower gain or increase the distance from the microphone.
There is no single number for every microphone and workflow. Aim for clear speech with headroom, no clipping and a stable gap above the room noise, then follow the target of the app you will use.
Other apps may apply their own noise suppression, gain control, codec and network processing. Use this page to validate the input path, not to predict every downstream effect.
Yes when the browser and operating system expose it as an input device. Bluetooth profiles can reduce quality or add latency.
Stop the test and confirm the browser’s recording indicator turns off. The implementation should stop every active media track when the page or test closes.